Dear Gerta, In relation to the Cis overlapping. I have a diagnostic test accuracy meta-analysis with the primary studies using small sample sizes. Therefore the confidence intervals for sensitivity and specificity are wide and all 10 studies overlap. This within study variability seems to mask any between study heterogeneity. Is it possible that a lack of precision in primary studies can make it look like the meta-analysis has low heterogeneity? James Meyer DVM MSc Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10
[R-meta] Sample size and continuity correction
2 messages · James Meyer, Röver, Christian
Dear all, just a brief remark -- I think the issue of overlapping / non- overlapping CIs is closely related to issues of discussing I-squared values, because you are essentially relating CI widths to differences between estimates. The cases of great overlap should be those where also I^2 is small, and vice versa. The conclusions you can possibly draw from looking at mere overlaps should hence have the same limitations as considerations of I^2 values. See e.g. the discussions by Ruecker et al. (2008; https://doi.org/10.1186/1471-2288-8-79) and Borenstein et al (2017; https://doi.org/10.1002/jrsm.1230). Cheers, Christian
On Thu, 2020-08-27 at 21:40 +0000, James Meyer wrote:
Dear Gerta, In relation to the Cis overlapping. I have a diagnostic test accuracy meta-analysis with the primary studies using small sample sizes. Therefore the confidence intervals for sensitivity and specificity are wide and all 10 studies overlap. This within study variability seems to mask any between study heterogeneity. Is it possible that a lack of precision in primary studies can make it look like the meta- analysis has low heterogeneity? James Meyer DVM MSc Sent from Mail<https://go.microsoft.com/fwlink/?LinkId=550986> for Windows 10 [[alternative HTML version deleted]]
_______________________________________________ R-sig-meta-analysis mailing list R-sig-meta-analysis at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis